Being able to test the difference that Shaders make isn't something that we normally test or get the chance to test very often. The launch of the GTX 480, though, was full of so much drama in relation to yields and its performance.
The biggest change that seemed to happen right in front of our eyes was a shift from the 512 Shaders we thought the card would ship with, to 480 Shaders. Sure, we knew it was going to cause a performance hit, but how much of a performance hit? The decreased amount of Shaders was only one of the issues with the GTX 480, though, with other issues being heat, noise and power draw.
While the latter didn't bother me, the first two were real issues; first we saw Galaxy attack the heat and noise with an awesome looking triple fan, triple slot cooler. We then saw GIGABYTE and MSI attack the model; in the end the GTX 480 looked like a great product and to be honest it still is, especially with some of the bargain prices it can be grabbed for now from some places.
We wonder, though, what would've happened if the GTX 580 we looked at today launched as the GTX 480 in March; the same clocks, but the new cooler and more importantly the 512 Shaders that we had hoped to have initially.
Something a bit different this time is the fact that we need to underclock our card here. Before we talk about that, let's cover the similarities. Both cards carry 48 ROPs, 1536MB of GDDR5 on a 384Bit bus, and on a 40nm core.
The two big differences are the Shaders and the clocks. The GTX 480 offers 480 Shaders while the GTX 580 offers 512; a number that we always had hoped the GTX 480 would carry.
The other difference is the clock speeds. The core clock on the GTX 580 is 772MHz verse 701MHz; this makes the Shaders 1544MHz and 1401MHz. As for the memory, the difference there is 4008MHz QDR verse 3696MHz QDR.
So, with Afterburner we've pushed our GTX 580 clocks down to come in line with the GTX 480. This will let us know just what those extra 32 Shaders do for performance without any other factors coming into play.